encode_submission/scan_extension.py

   1 from optparse import OptionParser
   2 import os
   3 import sys
   4 from pprint import pprint
   5
   6 def main(cmdline=None):
   7     parser = make_parser()
   8     opts, args = parser.parse_args(cmdline)
   9
  10     extensions = scan(args)
  11     common_extensions = find_common_suffix(extensions)
  12
  13     if opts.rdf:
  14         print_rdf(common_extensions)
  15     else:
  16         print common_extensions
  17
  18 def make_parser():
  19     parser = OptionParser("%prog: directory [directory...]")
  20     parser.add_option('--rdf', action="store_true", default=False,
  21                       help="Produce rdf configuration file for ucsc_gather")
  22     return parser
  23
  24 def scan(toscan):
  25     index = {}
  26     for cur_scan_dir in toscan:
  27         for path, dirnames, filenames in os.walk(cur_scan_dir):
  28             for filename in filenames:
  29                 base, ext = os.path.splitext(filename)
  30                 if ext in ('.daf', 'ddf'):
  31                     continue
  32                 next_index = index
  33                 for c in filename[::-1]:
  34                     next_index = next_index.setdefault(c, {})
  35     return index
  36
  37 def find_common_suffix(index, tail=[]):
  38     if len(tail) > 0 and len(index) > 1:
  39         return "".join(tail[::-1])
  40
  41     results = []
  42     for key, choice in index.items():
  43         r = find_common_suffix(choice, tail+[key])
  44         if r is not None:
  45             results.append (r)
  46
  47     if len(results) == 0:
  48         return None
  49     elif len(results) == 1:
  50         return results[0]
  51     else:
  52         return results
  53
  54 def print_rdf(common_extensions):
  55     import RDF
  56     from htsworkflow.util import rdfhelp
  57     model = rdfhelp.get_model()
  58
  59     viewName = 'http://jumpgate.caltech.edu/wiki/SubmissionsLog/NAME/view/'
  60     subView = RDF.NS(viewName)
  61     fileReTerm = rdfhelp.dafTermOntology['filename_re']
  62
  63     count = 1
  64     for ext in common_extensions:
  65         s = RDF.Statement(subView['VIEW{0}'.format(count)],
  66                           fileReTerm,
  67                           '.*{0}$'.format(ext.replace('.', '\\.')))
  68         model.add_statement(s)
  69         count += 1
  70
  71     writer = rdfhelp.get_serializer()
  72     writer.set_namespace('thisSubmissionView', subView._prefix)
  73     print writer.serialize_model_to_string(model)
  74
  75 if __name__ == "__main__":
  76     main()